Consulting on Quantitative Genetics: How Can This Work in My Breeding Program?

The Context of Canine Genetic Analysis

People and dogs have been doing things together for thousands of years: hunting, tracking, retrieving, and otherwise cooperating with people to obtain food and have fun. Over the last hundred years or so, human and canine cooperative work has been organized into more specific tasks. These are highly structured; some focusing on competition and others focusing on work like bomb-sniffing, patrolling, or alerting deaf people to sounds. Many of these activities take advantage of the dog's ability to perceive things that we can't, like odors. Since the attacks of September 11, there has been a renewed enthusiasm for working dogs, because they can help us in many ways, like detecting explosives, tracking a crime suspect, or search and rescue.

Over the last ten years or so, advances in computing power have enabled us to study animals' genetic complexity in more detail than ever before. On our desktops, we can now predict changes in quantitative genetic traits in future generations, using complex statistical models that take all the animals in the pedigree into account. We can study several traits at the same time, and how they are related to each other. We can collect lots of information about our populations because data has become relatively cheap to store.

In the years since September 11, there has been a worldwide shortage of healthy dogs with the potential to be trained or used in breeding programs. Organizations like guide dog schools, service dog schools, and government security agencies are coming to the conclusion that the best way to get dogs with predictable traits is to breed their own. The best way to use knowledge about the traits in a population to produce predictable stock and improve it over time is to use our knowledge of population genetics (the study of inheritance across a whole population) and quantitative genetics (the genetics of traits that vary on a measurable continuum).

Most of the traits of interest to us in dog breeding are quantitative -- rather than taking on discrete values, they are measured by numbers on a scale. PennHip scores range from 0 to 1, and OFA scores are also expressed in a range. Most parameters of structure, health, and behavior are a complex combination of genetic and environmental factors. With enough data on the animals in a pedigree, it's possible to begin to assess whether traits are heritable or to what extent they are heritable. We can also predict the likely results of selection in the future, which enables us to compare the possible results of different selection programs.

How We Can Apply Genetics to Practical Breeding Programs

There are three phases involved in researching and implementing a genetically-based breeding program in a population: the initial evaluation, ongoing study, and implementation. To maintain the breeding program, the implementation phase with periodic ongoing study to evaluate progress. Each of these require substantial collaboration with handlers, trainers, vets, and others with an interest in the dogs' work, health, and well-being.

Evaluation

The first part of a project would involve getting a sense of the data available. This would involve a description of the database rather than specific values. The important questions are:

• How many individuals are in the pedigree? Is there a lot of missing data in the pedigree? Is there data for all offspring, or just a few from each litter?
• What traits have you collected data on? How is this data stored? What methods are used to collect the data? How frequently are the data collected?
• How much inbreeding is there in the population? Where did the original dogs come from?
• Things the geneticist should learn from the experts on the population:
  • What traits are most important?
  • What are the experiences of breeders with these traits?
  • Do any of them seem correlated with each other?
  • How have selection decisions been made in the past?

Ongoing Study

The study phase would involve actually analyzing the data statistically to learn about the population size, structure, and level of inbreeding. The heritability of each trait should be estimated, and correletions between the important traits should be estimated.

The results of these studies should be discussed in detail with the people interested in the breeding program, and a collaborative decision made about which traits to focus on and what methodology to use for selection.

Implementation

After priorities are set, a program of selection will be implemented. This program will be used to select parents for the next generation or generations until the population is re-evaluated. This usually involves using BLUP or a selection index to choose animals with the best values of the important traits while minimizing inbreeding. After a few generations, it is important to return to the evaluation phase to ensure that the desired results are obtained. The three steps in applying genetics to a breeding program form a cycle. Periodic analysis and discussion of priorities keeps the focus of genetic change on the important traits.

Questions

Why is it important to return to the evaluation phase often?

Heritability is a number between 0 and 1 that represents the proportion of the variation in a trait that is due to genetic factors. If I say that the heritability of a trait is 0.40, it means that genetic factors account for 40% of the variation, and environmental factors account for 60% or the variation. These statistics apply only to the population in which they are measured. Each population has a different combination of genetic and environmental factors. A litter of puppies that goes to many diverse homes may have much more environmental variation in some trait than a litter that is raised all together, say, for the first 3 months of life.

As time passes, genetic parameters like heritability and genetic correlations even change within the same population. We usually expect that as we continue to select for a particular trait, we decrease the genetic proportion of the variation. By selecting, we are choosing breeding dogs who have similar values for the trait under selection. We are not using all the genetic variation that exists in the population; we are only choosing the best. As the genetic variation decreases, the environmental variation increases-- the environment becomes more important in determining the value of the trait. We need to update our estimates of these parameters in order to accurately assess and predict the changes we want to make in the population.

Why are correlations between traits important?

When we look closely at quantitative traits, we find that they can be related to each other. An obvious example would be that the length of the long bones in the front legs is probably correlated with the length of the long bones in the rear legs. This would be a positive covariance; as one trait increases, so does the other. In dairy cattle genetics, we have an example of a negative genetic covariance: there is a slight negative covariance between milk production and fertility. As we have selected for increased milk production, fertility has decreased a small amount.

Because of the possible unexpected genetic covariances between traits, it's important to evaluate all the important traits in the population to ensure that applying selection pressure to one won't cause unwanted changes in another.

Such covariances can occur for many reasons. They can be due to the phenotypes of both traits being dependent on the same biochemical pathway, like growth regulation. Behavioral or working traits could be correlated because they depend on similar underlying abilities.